Opinion Strategies for Political Organizations: Astroturfing, Influence Networks, and Shaping Narratives

Political topics are inherently divisive — for every supporter, there's an opponent. For political organizations pushing policy agendas or seeking public backing, decisions often require choosing the least damaging path between bad options. But the practical difficulty of any given push is largely shaped by public opinion — and public opinion doesn't form entirely on its own.

The Reality of Public Opinion: Echo Chambers and One-Sided Information

In practice, the proportion of people who actively seek out the full picture and verify facts independently is quite small. Most people receive information within their own echo chambers and form positions based on what the people around them broadly believe.

This implies two things:

1. The importance of echo chamber depth: A thicker, more concentrated echo chamber means more supporters, greater volume, and stronger outward influence.
2. Disrupting the opposing echo chamber: Strategically weakening the cohesion of the opposition — making their voices scattered and self-contradictory — is a common tactic in opinion management.

Coordinated Networks: The Core Engine of Narrative Influence

In political operations, two types of coordinated online forces are commonly deployed.

• Organized networks (网军): Systematically publish position-aligned content across social platforms, forums, and comment sections — using account volume and posting density to steer the direction of discussion.
• Surge networks (水军): More agile than organized networks, capable of flooding specific discussion areas with likes, comments, and shares in a short time to rapidly shift the atmosphere.

Both types share the same core objective — influencing collective perception. Whether making supporters appear to be the majority, or making opposition voices seem disorganized and fringe, the goal is always to tilt neutral or undecided audiences toward your side.

Common Narrative Management Strategies

Four tactics are commonly observed in practice. Understanding how they work also helps in recognising the traces they leave behind in the information environment.

1. Concentration of resources at critical junctures When a specific event breaks, a policy is announced, or a controversy is just beginning, operators pour resources in quickly to dominate the initial discussion space. First impressions in the information environment often shape how the story is perceived long-term.

2. Manufactured multi-dimensional support Support for a position does not stay in a single script. The same viewpoint is presented from different angles, different personas, and different types of accounts, making it feel more genuine and organic. A burst of closely worded posts from accounts claiming otherwise unrelated identities is the usual trace of this tactic.

3. Undermining opposition narratives Against opposing arguments, confusion, topic deflection, or divisive secondary controversies are introduced to splinter what was a unified opposition into conflicting factions.

4. Data monitoring with real-time adjustment Social listening tools track discussion volume and keyword trends across platforms, and resource allocation is adjusted according to what the data shows.

Why These Operations Persist

In an ideal democratic society, public discourse would be pluralistic and organic. But the reality is that once one side deploys these tactics, the other side relying solely on naturally occurring support is usually at a disadvantage at decisive moments — and that mutual catch-up pressure is the structural reason these operations persist across political arenas worldwide.

Persisting, however, is not the same as being lawful. The judgments and penalties set out later in this article show that once an operation involves fabricated content, concealed identity, or sock-puppet accounts, the Criminal Code, the Civil Servants Election and Recall Act, and the Fair Trade Act can all apply directly. The point of understanding how this works is to let readers see what is actually happening in the information arena.

A Reader's View: Spotting Discussion That Has Been Pushed

Coordinated networks, influence operations, and narrative management are a real feature of the modern information environment. For an ordinary reader, what matters is not how loud any of it gets, but being able to tell which discussions formed organically and which were pushed.

The clues are rarely in a single post — they are in the overall pattern: a concentrated influx of new accounts in a short window, closely worded comments from accounts claiming unrelated identities, and a topic focus that keeps getting deflected. The information arena has never been a level playing field, and recognising these signals is what keeps an amplified volume of posts from being mistaken for majority opinion.

The Risks of Overreach: Warnings and Limits

While coordinated networks can dramatically shift opinion in a short time, pushing too far can backfire and hand ammunition to the opposition. Numerous historical cases show that whether it's an individual, a political figure, or a corporation — once exposed for using fabricated information to manipulate opinion, the consequences typically include severe reputational damage, legal action, and fierce counter-narratives.

Real Case Review

1. 2018 Kansai Airport Typhoon Online Opinion Operation: Two defendants posted content on PTT denigrating Taiwan's Economic and Cultural Office in Osaka, and directed members of a LINE group to amplify it. The first-instance court sentenced each to 6 months for insulting a public institution; on appeal this was re-characterised as insulting a public official in the execution of duties, with 5 months each. The offence of insulting a public institution was later repealed by legislation, and the "insulting an official duty" limb was struck down by Constitutional Court Judgment 113-Hsien-Pan-5. On 14 February 2025 the Taiwan High Court entered a final judgment of exemption from prosecution, so neither defendant was ultimately subject to criminal punishment. (Taiwan High Court, 113-Hsien-Geng-Shang-Yi No. 3)
2. 2023–2024 Legislative Election Disinformation Case: During the 2023 campaign, false content insinuating misconduct by a sitting legislator was posted on Dcard. On 30 April 2026 the Taiwan High Court, on appeal, sentenced one defendant to 6 months' imprisonment under the Civil Servants Election and Recall Act for profit-motivated dissemination of false statements intended to cause a candidate to lose; the three co-defendants were all acquitted for insufficient evidence. (CNA report, 30 April 2026)
3. 2023–2024 Taoyuan District 6 Sock-Puppet Smear Case: A marketing firm used the LDPlayer emulator to mass-create Facebook sock-puppet accounts (144 are listed in the judgment's schedule) and posted attack comments and memes under the public posts of city councillors campaigning for a candidate, intending to cause a candidate in that district to lose. The Taiwan Taoyuan District Court convicted two defendants under Article 104(1) of the Civil Servants Election and Recall Act for disseminating false statements, sentencing them to 7 months (suspended for 3 years, with an order to pay NT$200,000 to the public treasury) and 6 months respectively, each with deprivation of civil rights for 1 year. (Taiwan Taoyuan District Court, 113-Yuan-Hsuan-Su No. 2)
4. 2013 Samsung "Ghost Writer" Scandal: Online writers were hired to conceal their commercial identity, pose as ordinary consumers and disparage a competitor's products. The Fair Trade Commission found this to be a "deceptive act sufficient to affect trading order" and imposed fines of NT$10 million on Samsung Electronics Taiwan plus NT$3 million and NT$50,000 on the two contracted agencies — NT$13.05 million in total. (Fair Trade Commission Decision No. 102184)

Primary Legal Risks

1. Criminal Code: Defamation (Article 310); public insult (Article 309 — narrowed by Constitutional Court Judgment 113-Hsien-Pan-3, which limits punishment to demeaning speech that, read in its expressive context, exceeds what an ordinary person can reasonably be expected to tolerate and remains unprotected after balancing; subjective feelings of honour are not protected); insulting a public official (Article 140 — narrowed by Constitutional Court Judgment 113-Hsien-Pan-5, which requires the subjective purpose of obstructing official duties and conduct clearly capable of disrupting the official's command, communication and performance of duties). The previously listed offence of "insulting a public institution" was repealed by legislation promulgated on 12 January 2022, and the "insulting an official duty" limb of the same article was declared unconstitutional and void from 24 May 2024. (Criminal Code Article 140, Laws & Regulations Database)
2. Personal Data Protection Act: Unlawfully obtaining, processing, or using others' personal data.
3. Fair Trade Act: Business defamation provisions.
4. Civil Servants Election and Recall Act, Article 104: Whoever, with intent to cause a candidate to be elected or not elected, spreads rumours or disseminates false statements by text, image, audio, video, speech or other means, in a manner sufficient to cause damage to the public or to others, faces up to five years' imprisonment; where deepfake audio, images or electromagnetic records are used, up to seven years; and where the offence is committed with intent to profit, the penalty is increased by up to one half, with a possible additional fine of NT$2 million to NT$10 million. (Civil Servants Election and Recall Act Article 104, Laws & Regulations Database)

The Risk Threshold and Strategic Red Lines

Opinion management is not a simple volume contest — it's a strategic game where risk and influence coexist. When technology and manpower can amplify your voice rapidly, errors and missteps will be amplified just as quickly and reflected back at you.

As the four cases above show, the risk threshold generally has little to do with how loud a campaign gets. It turns on whether the operation involves fabricated content, concealed commercial identity, large numbers of sock-puppet accounts, or demeaning speech aimed at an identifiable target.

The four cases did not end the same way, and that difference is itself the point. In cases 2 and 3 defendants received prison sentences under Article 104 of the Election and Recall Act, and in case 4 the Fair Trade Commission imposed fines — liability did attach. But the two defendants in case 1 ended up with the prosecution dismissed and no criminal penalty at all — not because the conduct was found harmless, but because one of the offences charged was repealed by the legislature and the other was struck down as unconstitutional.

So there are two lessons, not one. On conduct, the red lines are the items above. On law, the provisions those lines rest on do change — the same conduct can land very differently depending on when it is prosecuted.